Insulator Contamination Prediction Model Based on BP Neural Network Optimized by Genetic Algorithm

In order to predict insulator contamination degree on the transmission lines, this paper proposed an insulator contamination prediction model based on back-propagation (BP) neural network optimized by genetic algorithm. In view of the inadequacies of the slowness of convergence rate and the weakness of global optimization ability, the genetic algorithm was used to improve the performance of prediction result. This paper not only consider meteorological factor which includes temperature, wind, precipitation and relative humidity as input layer but also consider air quality index (AQI) which contains PM2.5, PM10 as input layer. Moreover, the equivalent salt deposit density (ESDD) and non-soluble deposit density (NSDD) were set as output variables. The results show that the optimized prediction model has better ability in searching optimization. Furthermore, the prediction result of proposed model is more accurate than that of the BP neural network.